Very insightful post by Seth Grimes identifies four missing measures:
- sentiment classified in ways that are meaningful to business, such as “promoter/detractor,” “satisfied/disappointed,” “happy/sad/angry” or whatever are relevant for your needs. Grimes points out that flexible automated methods, and expert or crowd-sourced analysis is up to the task. Netvibes, for example, offers a very sophisticated capability that allows in-house analysts to classify sentiments using any terms that make sense, such as “threat,” “opportunity,” “product improvement,” “unmet need,” and so on.
- sentiment density – does the post use a few words or is it packed with words that indicate a lot of feeling;
- variation – the dispersion of words around an idea; and
- volatility – a measure of the variation of sentiment over time. Seth isn’t saying that these measures are not being used today – they are, rather it’s that they are not used broadly enough. Too many sentiment analyses leave it at the positive/negative/neutral level.
I especially appreciated Seth’s final point, not a metric but an exploratory process called “root cause analysis.” That gets us to understanding the “why” underlying the “what” reported by your metrics and indicators.